Slightly More Chaos, Slightly Less Control
Mid-year reflections from a PM trying to figure it out
We’re halfway through 2025, and usually around this time of year I like to do a bit of self-reflection: am I where I wanted to be 6 months into a year that’s seemingly flying by?
This year’s halfway check in gave me a fright. Don’t get me wrong, there’s always things I could do better, or goals/resolutions I just ignored that need some TLC. But this year, I’ve come to the somewhat sobering realisation that I’m behind the curve, and the curve has changed. I was busy, tried to be strategic, on top of my roadmap and had my vision statement at the ready. But if I’m honest, beneath it all has been an undercurrent of mild panic, quick fire learning and a quiet “ah, that’s new” at least once a day.
There’s a shifting industry, new technologies, new expectations on our role and an honestly loud, sometimes overwhelming thought-leadership pool. There’s been so much to stay on top of, it’s at times felt like I’m at the bottom of that San Francisco hill in the Sony ad with the bouncy balls – and it’s my job to try and catch them.
So here it is – my oh no list for 2025.
1. I thought AI would be someone else’s job
Ok this one is cheating. Not necessarily a realisation I’ve just made, I think the naive optimism of AI not being a PM problem died a long time ago – I knew I’d have to face the curve eventually and try to find my space on it. But if I’m honest, the acceptance was followed by paralysis - where do I start?
The reality check
It’s everywhere, and it’s a growing requirement for the job. You're probably not going to write your own LLM, or build your own Gen AI tool, but to be comfortable in the noise of the technology and be able to look for opportunities to bring it into your fold, that’s on you.
If your job is to help find those chances for growth, opportunities to delight, solve problems your users didn’t know they had – then knowing how, when and why an AI tool could help you is important. Knowing its benefits - how it could accelerate solving a problem for you, and its pitfalls - how it could hallucinate its way into a nightmare, will help you ask the right questions.
Lessons
AI isn’t a feature. Think of it as a helpful colleague that’s sometimes unpredictable, and you’re their work buddy. You’d better get to know them well.
I’ve spent more time over the last 6 months trying to retroactively get a grip. On AI, but also myself. Reading, asking questions, listening to people ahead of the curve to me. I was going to say better late than never, but I need to remind myself (and you) that we’re still so early in the lifecycle. It’s ok to feel overwhelmed, but any step onto the learning curve is a step toward progress.
Here’s some questions to help you get started:
What kind of models exist?
How are they different?
What problems could they solve? How do we know they’re solving them?
And if you’re working on a product using AI for the first time:
How will we know if the model is wrong?
How do we clearly signpost this is AI?
How do we build is graceful failure modes?
2. Alignment isn’t approval
A difficult one to admit, to be honest. I’m a people pleaser, and I get a sense of satisfaction from approval. Unfortunately for me, that slightly goes against what I need to be a good PM.
Don’t get me wrong, that isn’t to say that bringing people along the journey, and getting their input is wrong. That is 100% the goal here. But I think where I (and probably some of you) struggle is the decision point. Getting 100% of the room agreed on a decision is:
Unrealistic
Time consuming
Costly
Reality check
Not everyone will have 100% of the context, 100% of the time. Your job is to share as much as possible, gather as many of the relevant insights as you can, ask the right questions, tell the story of the problem as best you can, and collect the relevant data points that will help you decide on a solution. Objectivity is the goal, but subjectivity is your final boss. When you have a room full of stakeholders, the fact is, some will hear, see and understand the narrative. Some will have unbacked opinions. If you’re set on 100% alignment, then what you’re really looking for is approval.
Lessons
PMs aren’t approval seekers, they’re clarity bringers. And whilst in theory I’ve always known this, learning how to phrase this with confidence, and navigate the push back, has been a lesson long in the making. FYI, it’s something like this, but said with the confidence of a data-backed decision:
“ We’re going this way. Here’s why. I know not everyone agrees, and that’s OK”
Lead with the Why
Frame your decision around your shared goals. Recontextualising around the bigger pictures helps people move away from the tactics
Expect the DM
The quiet “Hey, can we talk about this decision” message always comes. It’s an invitation to clarify and build trust.
What’s your biggest concern? What trade-off do you think we didn’t consider?
Sometimes, it will uncover a blind spot and save you from a wrong decision. Often, it will show you’ve already considered it thoughtfully. Either way, trust builds.
Stay calm if it gets frosty
If the tone gets cold, resist the urge to defend emotionally (my pitfall to be honest). Used structured reasoning and shared language:
“Here’s the criteria we used”
“Here’s what changed since our last conversation”
“Here’s how we’ll know if we were wrong, and how we’ll course correct”
The goal isn’t to win the argument, but to maintain credibility when things get tough.
Loop them back in
Make sure you invite them back in to follow up conversations
People are less likely to fight your decisions if they feel part of the next one.
3. Control is a lie, but that’s OK
Product Management has never been about control, but admitting you lack control is scary. Adaptability’s always been a core trait for us, but I think this year it’s taken centre stage.
Reality check
We deal with complex & fluid systems, try to influence without ownership, [try to] stay calm in the face of AI and a volatile industry, managing constant and sometimes growing tension between your stakeholder, your market and your user.
Lesson
Relinquishing the idea that you can control all of that, all of the time, and focusing more on where your time, brain and energy are most useful in that moment – that'll take you far.
Ultimately
It’s been an odd but fun 6 months, with imposter syndrome weighing heavier than usual, but from conversations with friends and colleagues, probably not an entirely unique 6 months.
We’re still here, still building, still learning.
And I think if there’s anything I’m going to take from this personally, it’s that staying curious beats staying in control – every time.
 | A guest post by
|